Is anyone here familiar with the proof (using homology) of the generalized Jordan curve theorem, that a subspace of S^n homeomorphic to S^(n-1) divides it into two components? It can be found on page 169 of Hatcher's algebraic topology book, which can be downloaded from this page. I'm wondering if it's possible to use the same kind of proof to show that a general (n-1)-manifold divides S^n into two components. It was shown that the complement of an n-1-sphere in S^n actually has the homology of S^0, which is much stronger, and won't hold for general n-1-manifolds. But all I need is that the 0th reduced homology group is Z, which does seem to be true. The problem is that the corresponding terms of the Mayer-vietoris sequence used in Hatcher's proof aren't zero where they need to be. Does anyone have any ideas?